Create images under new namespace
rndmh3ro opened this issue · 5 comments
Is your feature request related to a problem? Please describe.
Now that the docker images are moved to this new organization, we should create the docker images under the new organization, too. We need to check which registry we will use.
If the Dockerhub provides a redirection-method when you docker pull
and image, we should use that. Otherwise we should push to the old namespace as well as the new one.
IIRC redirects are not possible due to how image paths are stored in the manifests themselves.
But @rndmh3ro IMO it might make sense to push directly to ghcr.io as there is no need for static credentials and you avoid any further changes to Docker Hub limits/pricing & policies that have been going on recently.
I think changing to ghcr.io would simply be a matter of changing the workflow files and removing the static secrets, and once they start building on main the new images should appear in this project's registry.
The old images would stay on dockerhub so it wouldn't break anything until downstream references are updated to use the new registry.
/cc @dlouzan
I am also in favor of using ghcr.io instead of Dockerhub
FWIW I just tried ghcr.io on the all.yml
workflow to test a manual push and it looks very smooth:
7af061a...2922f5c
https://github.com/nejch?tab=packages&repo_name=docker-ansible
https://github.com/nejch/docker-ansible/pkgs/container/docker-alpine-ansible
I can open a PR later (with cleanup and all the other changes needed) if this make sense to you.
I can open a PR later (with cleanup and all the other changes needed) if this make sense to you.
Yes, that'd be great. However I'd like to push the images also to the dockerhub in the rndmh3ro-namespace. Can you add both, please?
I can open a PR later (with cleanup and all the other changes needed) if this make sense to you.
Yes, that'd be great. However I'd like to push the images also to the dockerhub in the rndmh3ro-namespace. Can you add both, please?
Sounds good, will do! I'll wait for #22 to make sure that's deployed correctly first.